The channel post worked. Someone tapped Buy. Then the button opened a website, and the order died in a browser they did not ask to open.
That hop has a name in merchant slang: link-out. It is a buy button whose next screen is not Telegram. A Telegram payment bot does the opposite. The invoice, the total, and the charge stay in the chat the buyer already has open.
This is the conversion argument, not another setup tutorial. For provider tokens, Stars, and Stripe, use how to accept payments in Telegram. For which rail belongs on a parcel versus a file, use Stripe for parcels, Stars for digital.
The buyer left the trip, not the product#
Baymard Institute's average of 50 published studies is a 70.22% cart abandonment rate, updated 22 September 2025. About 30 of every 100 people who reach a normal web checkout finish it. That number already assumes they arrived.
Link-out adds a journey before that checkout exists. Baymard never counts the person who closed Telegram's built-in browser on the first paint. They never had a cart.
Among people who do abandon, and ignoring the 42% who say they were only browsing, the reasons that a website handoff makes worse are specific. Shoppers could pick more than one, so these shares overlap and must not be added:
| Reason they gave up | Share of abandoners | What a link out of Telegram does |
|---|---|---|
| Extra costs too high (shipping, tax, fees) | 40% | Unchanged. A Mini App with a surprise courier fee fails the same way. |
| Did not trust the site with a card | 19% | A new domain, inside a webview, with no Telegram identity on it. |
| Site demanded an account | 18% | The website does not know the Telegram user who just tapped Buy. |
| Checkout was too long or complicated | 17% | A second site, a second form, often a second login. |
| Could not see the total up front | 12% | The channel said a price. The website adds shipping after the hop. |
| Not enough payment methods | 9% | Apple Pay and Google Pay often never render in a messenger's built-in browser. |
Baymard also finds that an average large store can still lift conversion 35.26% by fixing checkout design alone, on sites that already have a normal browser. Telegram sellers who export checkout give that problem a head start.
What the link actually opens#
A t.me post that points at yourstore.com/checkout does not open Safari or Chrome. It opens Telegram's in-app browser. That webview is a bad place to pay:
- No Telegram identity. The Mini App container can pass the signed-in user into your store. A random URL cannot. The buyer is a stranger on a new domain, so the site asks for an email, a password, or a phone.
- Wallets go missing. Apple Pay and Google Pay depend on a real browser context. Inside a messenger webview they often do not appear at all. The buyer is left with a card form, then a 3-D Secure redirect that the webview mishandles.
- The back button deletes the session. Returning to the chat does not return to the cart. The cart lived on the other origin, in a cookie the in-app browser may not keep.
- Trust resets. The channel was a person they follow. The checkout is a domain they have never paid. Baymard's 19% "I didn't trust the site with my card" is this moment.
A payment bot skips that browser. Telegram's Bot Payments API sends an invoice. The buyer confirms a native payment sheet: card provider for physical goods, Telegram Stars for digital goods. The receipt comes back into the same chat.
Three bots that look alike and convert differently#
Search results for "telegram shop bot" and "telegram payment bot" mix three products. Only one of them sells without a link-out.
| What you ship | What the buyer does | Where the money is taken |
|---|---|---|
| Link bot | Taps a URL, lands on a website | The website's checkout, inside Telegram's browser |
| Payment bot | Confirms an invoice in the chat | Telegram's payment sheet. No catalog. |
| Shop bot (Mini App) | Browses, carts, pays in one window | Same payment sheet, after a catalog that never left Telegram |
A link bot is a bookmark. It can be useful for a menu, a size chart, or a policy. It is a bad Buy button.
A payment bot is the right tool when the offer is one product and one price: a consultation, a drop, a fixed bundle. You do not need a storefront. You need sendInvoice (or the Mini App equivalent, openInvoice) and a provider token from @BotFather.
A shop bot is the right tool when people choose. Variants, delivery, promo codes, and a cart belong in a Mini App, still inside Telegram, still ending on that same sheet. Tiny Shops is that shop: a no-code Telegram shop builder with the catalog and the invoice in one Mini App. Tiny Shops takes 0% on the charge. Stripe's fee, or Stars' fee, is still yours to pay.
The same 100 taps, counted without a fake rate#
Start with 100 people who tap Buy on a channel post. Give the website the benefit of the doubt and assume all 100 reach its checkout. At Baymard's 70.22% abandonment, about 30 orders complete. That is the ceiling, not the forecast. Anyone who bailed in the in-app browser never entered the 100.
There is no published, Baymard-style average for Telegram-native checkout, so this article will not invent one. What the in-chat path removes is the rows in the table above that are about the handoff: the new account, the untrusted domain, the long second form, the wallet that never drew. What it does not remove is the offer. If 40% of abandoners cite shipping, tax, or fees, a Mini App that hides the courier until the last tap will lose them too.
Put the full total on the invoice before the pay button. Shipping, tax, and the item price in one number. The channel price and the sheet price have to match.
What staying in Telegram will not save#
Fix the rail after the offer is honest.
- A slow or expensive courier still loses the 40% and the 20% who said delivery was too slow. In-chat checkout does not invent a cheaper carrier.
- A vague returns policy still loses the people Baymard measured at 13%. Write it in the product screen, in the language of the channel.
- Digital goods on a card rail inside the Mini App are a compliance problem, not a conversion trick. Telegram's payments docs say the app can be hidden from mobile users if digital goods are charged in a normal currency through a third-party provider. Parcels use a card provider. Files, memberships, and credits use Stars. Mixing them on one Mini App is how shops disappear from iPhone. The full rule is in Stripe versus Stars.
A website link does not dodge that rule if the sale still happens inside the Mini App. It only trains the audience to leave.
When a link out is the right product#
Send people to the web when the web is the product.
- The buyer must configure something that does not fit a phone sheet: a made-to-measure quote, a B2B net-terms order, a desktop-only tool.
- The page is not a checkout. A lookbook, a size guide, a legal page, a long review. Those links do not ask for a card.
- You are deliberately selling on your own domain for SEO or wholesale, and Telegram is only the announcement. Then the button should say "Open the site", not "Buy". The label is the contract.
If the button says Buy, the next screen should be able to take the money.
Put Buy on the payment, not on a URL#
For a store that already has an audience in Telegram:
- One primary button on the post. It opens the shop Mini App or a single invoice. A second button for "also our website" splits the tap.
- The first screen shows the price the post promised, with shipping in the total before pay.
- Do not ask for an email to create an account. You can already message this person in Telegram. Ask for a shipping address only when a parcel needs one.
- Card, Apple Pay, and Google Pay for physical goods, through Telegram Payments. Stars for digital goods. The provider setup is the payments guide.
- After payment, the receipt and the shipping update stay in the chat. The buyer should not need the website to know the order is real.
That is the whole case. The audience is already in the app. The orders still available to you are the ones you stop exporting to a checkout that abandons seven out of ten people who reach it — and more who never do.
